The aim of the paper is to show several case histories regarding the diagnosis of rotor asymmetries and other faults in large induction motors. These are motors under real operation in power generating installations. Two different approaches are used for the diagnosis of the machine condition. Both of them are based on the study of the startup transient of the machine and the extraction of characteristic patterns caused by the evolution during the transient of certain components dependant on the slip. The results show the validity of both diagnosis methods. Special interest has the use of the Discrete Wavelet Transform for the detection and discrimination of different faults. Its validity was already proved by the authors in previous works but in small motors. The results show the potentiality of the approaches for the diagnosis of the condition of a wide range of motors.